Ask ten seniors how they picked their thesis topic and nine will tell you they waited for the right idea to show up. The tenth will describe a process. The tenth one hands in on time.
Picking a topic for your thesis or capstone project works in three moves. First you gather ideas broadly. Then you shrink every idea along four dials, time frame, place, target group and angle, until a single research question is left. Last you test whether that question can be answered with the data and the weeks you actually have. The result is a topic your advisor signs off on and that you can finish before the deadline.
📌 The short version
- One research question is the mark of a good topic.
- Four dials shrink it: time frame, place, group, angle.
- A credit hour is defined as a minimum, not a total.
- Your IRB decides what counts as exempt research, not you.
- Without real access to data, no empirical topic holds up.
Create a survey for free
With empirio.ai you can create a modern online survey in minutes — free to start.
- AI-built survey
- Adjust by drag & drop
- Real-time analysis
What makes a good thesis topic
A good thesis topic satisfies four conditions at the same time: you stay curious about it for months, it belongs to your major, it converts into a single research question and it fits the time you are given.
Students usually test only the first one. Curiosity does matter most, since a topic that bores you by week four will not carry you through the remaining two months. Curiosity alone falls short, though. Nobody grades you on how exciting your field is. Graders look at whether one question was answered cleanly.
What decides the outcome is how the four conditions work together. A topic you love that yields no answerable question becomes an essay. A topic with a sharp question and no path to data becomes an emergency fix three weeks before the due date.
| Condition | How to spot it | What happens without it |
|---|---|---|
| Curiosity | you keep reading on your own | momentum dies around week four |
| Fit with your major | a course in your program covers it | no advisor feels responsible |
| Answerability | you can state one single question | the paper becomes a stack of material |
| Feasibility | data and time both exist | the method gets swapped near the deadline |
Whether your thesis needs data of your own or can run on published literature is a call worth making right here. The tradeoffs and what they cost you in hours are laid out in the guide on the choice between an empirical and a theoretical thesis.
Step 1: gather ideas until you have something to choose from
Five places reliably produce topic ideas: the courses that came easily to you, the limitations sections of recent journal articles, an internship or a part-time job, the research your faculty are running and ready-made topic lists.
Limitations sections are the most overlooked of the five. Nearly every study closes with a paragraph where the authors spell out what their work could not do and what someone ought to study next. Researchers have written your research gap for you in full sentences. Ten minutes of reading buys you that.
Gather broadly here and judge nothing yet. Ten to fifteen ideas make a workable pool and you will strike out most of them.
- Courses and term papers: subjects you already have a head start on.
- Limitations and future research: open questions at the end of recent studies.
- Internship or part-time job: problems you ran into on the job.
- Faculty research: projects and specialties in your department.
- Topic lists: prompts to build on, not to copy.
Starting from a ready-made selection works too. Sorted by field, there are topic suggestions for bachelor’s and master’s theses to pick from. This article walks the road from idea to research question, that list hands you the raw material. If you plan to write at a company, the specifics of a thesis written at a company are worth reading, since the topic often comes with the placement.
Track ideas and questions in a table
A table works better than a list at this stage, because three columns show you instantly which idea already produces a question and which does not. Column three waits until step 3.
| Topic | Possible research question | Feasibility notes |
|---|---|---|
| Support in online courses | How do students rate the support they get online? | survey, reach through student government |
| Campus food pantries | What keeps eligible students from using the pantry? | needs the dean of students on board |
When an hour of searching turns up no question for an idea, strike it. Column two exists to make that call for you.
Our reading tip
Booth, Wayne C. et al. (2024): The Craft of Research. 5th edition, Chicago: University of Chicago Press. Its section on moving from topics to questions covers exactly the step described here and is the model for the table above.
Step 2: shrink the topic to one research question
Shrinking a topic means turning four dials one at a time: time frame, place or institution, target group and angle. “Online learning” becomes a question you can state in one sentence and answer with data you collect yourself.
Almost everyone shrinks in a single jump, which is the mistake. “Online learning” goes straight to “online courses at community colleges” and the target group, the time frame and the perspective are all still open. Turn one dial, then look at whether the topic still holds.
One test closes this step: you are finished when you can no longer state your topic in three words without dropping something that matters.
| Dial | Question to ask | Example |
|---|---|---|
| Time frame | Which stretch of time? | since the courses moved online for good |
| Place or institution | Where exactly? | community colleges in Arizona |
| Target group | Who are you looking at? | students in their first year |
| Angle | From which perspective? | how supported they feel, not their grades |
Put all four together and a question like this comes out: “How do first-year students at community colleges in Arizona rate the support they receive in online courses?” Not a pretty sentence, but one you can work with. Workable beats pretty at this point.
As long as three words cover your topic, you do not have a topic yet.
Turning that narrowed topic into a clean question is laid out step by step in the guide on how to write a research question. The final title of your thesis is a third formulation again and follows rules of its own: wording the title of a bachelor’s or master’s thesis.
Step 3: check feasibility before your topic is approved
A topic is feasible when you can answer its research question with data you can actually get within the weeks you have. Credit hours give you a first sense of the workload.
Federal regulation defines a semester credit hour as approximately not less than one hour of classroom or direct faculty instruction plus a minimum of two hours of out-of-class work each week for roughly fifteen weeks. The wording sits in 34 CFR 600.2. Multiply that out and you land at about 45 hours per credit hour, which is a floor derived from the minimum and not an official total.
No federal rule sets how many credit hours a thesis or capstone project carries. Your program handbook and your graduate school policies decide that, along with how many weeks you get. Read both before you commit.
IRB review decides what counts as exempt, not you
Your university’s Institutional Review Board decides whether your survey needs review and whether it qualifies as exempt. Students are generally not allowed to make that call themselves.
The Common Rule applies, as a matter of federal law, to research conducted, supported or otherwise subject to regulation by a federal agency, which is set out in 45 CFR 46.101(a). A student survey without federal funding does not fall under it on that basis alone. Your university most likely holds a Federalwide Assurance and many institutions extend that commitment voluntarily to all research with human subjects, which is how the rules reach your thesis anyway. Survey and interview research can be exempt under 45 CFR 46.104(d)(2) when the conditions are met.
Do not self-certify
Federal guidance steers institutions away from letting researchers decide on their own that their own study is exempt. Submit the form and let the IRB make the determination, even when you are sure of the answer. Details are in the OHRP guidance on exempt research determinations.
Access to the field breaks more empirical theses than statistics ever do. How many responses you need depends on what you plan to do with them, which the guide on the sample covers. Where to find respondents beyond your own contacts is covered in finding survey participants.
Checklist before you commit to a topic
Walk through these six points before your topic becomes binding. Any point you cannot answer with yes is an open item, not a detail that resolves itself later. Point four is where things stall most often.
- The research question fits in one sentence.
- Your advisor has said yes to the topic out loud.
- Every term in the question has literature behind it.
- Access to the field is settled, IRB approval included.
- The method follows from the question, not the reverse.
- Data collection and analysis fit your calendar.
Create a survey for free
With empirio.ai you can create a modern online survey in minutes — free to start.
- AI-built survey
- Adjust by drag & drop
- Real-time analysis
Finding a thesis topic with AI: what works and what does not
Language models are strong at spinning one broad field into twenty variations and weak at reporting the state of research accurately. Put them to work sorting and sharpening your own ideas, never as proof that a topic has enough literature behind it.
In practice the workflow looks like this. You hand over your field and your major and ask for narrower versions along the four dials from step 2. Nothing there replaces a real literature search, though it does get you past the blank page. Every idea that comes back gets tested the same way one of your own would.
Permission to use AI at all, and whether you have to disclose it, comes from your university. Plenty of program handbooks and academic integrity policies now spell this out, others stay silent. Ask your advisor before you start rather than when you sign the statement of original work.
Watch out
Language models invent citations that look convincing and lead nowhere. Check every title in your library catalog or a subject database before it reaches your proposal. A source you cannot find there does not exist for your thesis.
Finished tools for this step are around as well, among them the AI section of GoThesis, which belongs to the same company as empirio.ai. The rule from above carries over: suggestions are a starting point and the verification stays with you.
Common mistakes in choosing a topic
Topics rarely fail because they are bad. They fail because a step got skipped. Four patterns show up over and over.
Read the four sections below before you commit to a topic. After that point every correction eats time you need for writing.
The topic stays too broad for the schedule
“Artificial intelligence in human resources” sounds like a topic and is really a field with several thousand publications in it. Commit to that and your first weeks go into reading that never makes it into the paper, followed by narrowing under pressure. The usual result is a thesis that raises a lot and settles nothing.
The advisor gets asked too late
Plenty of students hunt for the perfect topic first and for someone to advise it second. Advising slots are limited and tied to specialties, so that order works against you. Talk to two or three faculty members early about a rough idea and let them tell you what fits their area. A no in week one costs nothing. A no in week four costs you the topic.
Data collection gets planned after approval
Late planning of data collection is the most expensive mistake and the hardest to see coming, because everything looks fine until the topic is approved. Only then does it turn out that the company will not allow a survey, that IRB review takes six weeks or that ten days produced fourteen responses. Settle access, approvals and your distribution plan while you are still choosing.
The whole topic gets traded in at the first setback
When data collection stalls, the first instinct is the biggest lever: a new topic. The fault almost never sits in the topic itself. One dial is usually all that needs adjusting. Shrink the target group, trade a survey for five interviews, move the time frame or limit yourself to one institution. Rescuing a topic takes days. A new topic costs weeks.
Conclusion
Choosing a topic comes down to three decisions rather than a moment of inspiration: gather, shrink, check. Anyone who pulls the third decision forward and settles where the data will come from before the topic is approved skips the worst experience of the whole degree. And once your topic fits in one sentence that ends with a question mark, you are ahead of most of your class!
What comes next
- Want the full sequence? Empirical research process: 7 steps with an example
- Writing your first thesis? Bachelor thesis: guide and structure with an example
- Need the section order? Thesis structure: the required order of every section
Topic settled, data still missing?
With empirio.ai, an online survey tool from Germany, you create the questionnaire for your thesis free of charge, share it as a link or QR code and get your responses analyzed as they come in.